Resection of the primary has its own risks. The risk of mortality with resection of the primary has been found to be between 1.6 and 4.6 % [19, 20]. Resection of the primary tumor delays initiation of systemic therapy, particularly if postoperative complications occur, potentially allowing progression of metastases. Rarely, patients may still be at risk for complications from the primary tumor, such as obstruction or perforation, if there is recurrence of the tumor either at an anastomosis or at another site in the bowel. In fact, a recent Cochrane review concluded there was no significant reduction in the risk of complications when surgery as the initial therapy was compared to chemotherapy and/or radiation as the initial therapy [21]. These findings have lead several authors to recommend avoidance of resection for asymptomatic primary tumors, [18, 20, 22, 23] while others recommend resection, particularly for patients who are low-risk for surgery from a medical standpoint [24, 25].

Resection of the primary could also theoretically slow the progression of the disease by debulking the tumor, allowing chemotherapy to work more effectively on the remaining disease. Many studies have examined the possibility of a survival benefit with resection of the primary tumor, but these have been hampered by the strong influence of patient selection since these were retrospective studies. Often patients who were in better medical condition or those who were thought to have a possibility of cure with from a combination of resection and chemotherapy were chosen for resection, leading to a bias toward much more advanced disease in the nonresected subjects. Results of such studies have varied widely. For studies limited to patients with asymptomatic primary tumors and unresectable metastatic disease, some studies have shown a survival benefit while the majority have not (Table 2). The authors of these studies have been split on whether resection of the primary is beneficial. Some have advocated for this approach [26] whereas others have not [20, 27–29]. If the criteria are expanded so that studies that include symptomatic primaries or resectable metastatic disease are included (Table 3), the picture becomes even more clouded. In these studies, it is difficult to determine how much of the improvement in survival, if demonstrated, could be due to patients who are curable. Some multivariate analyses have found tumor resection to be an independent predictor of survival [30–32] while others have not [33]. A Cochrane review on the subject concluded there is no consistent improvement in overall survival with resection of the primary tumor [21].

Thus, retrospective data have been very limited in their ability to allow reliable conclusions about the role of resection of the primary due to issues with patient selection. Ideally, a randomized controlled trial could be done among patients with asymptomatic primaries and unresectable metastatic disease who are medically fit to undergo resection, thus eliminating these patient selection factors that have plagued retrospective studies. However, recruitment for such studies has been difficult, with at least two studies being closed due to lack of recruitment [34, 35]. This difficulty with recruitment likely relates to patients being unwilling to undergo surgery for unclear benefit, or care providers encouraging them to have resection due to the perceived risk of complications from the primary. There is currently another randomized controlled trial on this subject underway, [36] which one hopes will provide definitive data as to whether primary tumor resection is associated with decreased complications or lengthened survival.

Other important issues in these cases are quality of life and patient preferences. Since surgery can be associated with short-term decreases in quality of life, this could be a major consideration in patients who likely have a limited life expectancy. Patient preferences in this area could be examined with the time-trade-off method, where patients are given a hypothetical scenario where the typical postoperative course for surgery is described in detail and patients are asked if an increased survival of 6 months, for example, would lead them to choose surgery and its associated recovery in order to gain this survival benefit. The anticipated increase in survival is then lessened and the same question asked again, with the process repeated until the patient no longer reports they would be willing to have surgery for the anticipated survival benefit. Such a study could provide information as to whether most patients would consider resection of the primary, even if it were proven to be associated with some definite increase in survival.

Thus, it remains unclear whether resection of the primary tumor is associated with a decrease in complications from the primary tumor or an increase in survival. Patient preferences in this area have not been well studied and are likely to play a significant role in whether surgical intervention for the primary tumor is pursued.